Abstract
Raman spectroscopy is a powerful method for analyzing chemical compositions across diverse samples. Spontaneous Raman scattering (spRS) provides complete Raman spectra but typically yields low signal levels, requiring long signal integration times. In contrast, stimulated Raman scattering (SRS) produces much stronger signals, allowing for rapid spectral acquisition, and has been widely used to accelerate chemical imaging. While theoretical comparisons under ideal conditions suggest that performance depends on photon flux and integration time, little experimental work has been conducted to systematically compare the limit of detection (LOD) of spRS and SRS spectroscopy. Here, we comprehensively compare the LOD of frequency-domain spRS and SRS for water-soluble analytes in aqueous solutions, a common environment for biological specimens. We introduce a LOD estimation methodology by taking only three measurements to derive the dilution maximum. This approach greatly simplifies the determination of the LOD compared to the conventional method of using serial dilution, thus enabling the convenient and rapid evaluation of various factors that impact the LOD in linear spectroscopy. Using this method, we assessed factors influencing the LOD of SRS spectroscopy and determined optimal conditions. At a short spectral acquisition time, SRS exhibits shot-noise-limited performance, whereas at a longer acquisition time, photothermal effects, such as stimulated Raman photothermal absorption, and low-frequency 1/f noise can cause deviations from ideal LOD scenarios. Under optimized experimental conditions, spectral-domain SRS using slow frequency sweeping confidently detects ∼700 μM dimethyl sulfoxide in water and 1 g/L glucose and protein in aqueous solutions. This performance can slightly surpass that of spRS spectroscopy with equal or even longer integration times.